Need help with the iDCT setting in mpeg2dec

The old Video Software Help forum, left visible as an archive.

Need help with the iDCT setting in mpeg2dec

Postby (NERD Studios) Arturo » Tue Nov 05, 2002 12:21 am

Hello everyone, I'm nit-picking over all the iDCT options for mpeg2dec presented in Ermac and AbsoluteDestiny's guide and I can't for the life of me figure out which option suits my processor. I know it probably won't make much difference, but I've been know to obsess about this kinda thing. Anyway, just to save everyone the trouble of looking them up, the options are:

>0 : Default value (in .d2v file)
>1 : 32 bit MMX
>2 : 32 bit SSEMMX
>3 : 64 bit FPU
>4 : 64 bit IEEE-1180 Reference
>5 : 32 bit SSE2 (for P4)

I have an Athlon XP 1800 processor and I would really appreciate any help that anyone could give me.
User avatar
(NERD Studios) Arturo
 
Joined: 19 Jul 2002
Location: Stillwater, OK

Postby klinky » Tue Nov 05, 2002 2:01 am

I would go with #2, since Athlon XP supports MMX + SSE, is not 64bits and does not support SSE2.


~klinky
User avatar
klinky
 
Joined: 23 Jul 2001
Location: Cookie College...

Postby (NERD Studios) Arturo » Tue Nov 05, 2002 2:52 am

I'll be sure to pick that option then. Thanks for your help. :D
User avatar
(NERD Studios) Arturo
 
Joined: 19 Jul 2002
Location: Stillwater, OK

Postby AbsoluteDestiny » Tue Nov 05, 2002 1:49 pm

You can just choose 0 for default, which is fine. The other options are just there for optimisations on faster processors.

I might add what processors can do what, but it really doesnt make *much* of a difference and l33t people will know what to choose :)
User avatar
AbsoluteDestiny
 
Joined: 15 Aug 2001
Location: Oxford, UK

Postby RadicalEd0 » Tue Nov 05, 2002 4:47 pm

dont forget that kwalitytards will always choose floating point reference iDCT because, well, it reconstructs the motion vectors more precisely than 32 bit mmx. However, there is barely any sort of difference at all and unless you have a really serious case of kwalitardation, you can leave it at default

oh yeah, I just remembered doom wrote an article on this
http://www.doom9.org/idct.htm
have fun
NMEAMV: PENIS
NMEAMV: IN
NMEAMV: YO
NMEAMV: MIXED
NMEAMV: DRINK
User avatar
RadicalEd0
 
Joined: 24 Jun 2002

Postby klinky » Tue Nov 05, 2002 10:46 pm

I never used any of those options yet :p. I just learned about that feature from this post. I do remember though way back when I was trying to steal rented dvds that enabling MMX for IDCT in FlaskMPEG caused really fuckin crappy looking encodes. I think it's just since Flask sucked really tho.


~klinky
User avatar
klinky
 
Joined: 23 Jul 2001
Location: Cookie College...

Postby Mask of Destiny » Tue Nov 05, 2002 10:52 pm

MMX uses integer arithmetic and as such isn't as accurate as Floating point. Generally speaking this doesn't cause enough error to be noticable. Check out the article RadicalEd0 posted.

I've used Flask once or twice and didn't have any quality problems. I've never tried using the MMX one though.
Mask of Destiny
 
Joined: 11 Aug 2002

Postby (NERD Studios) Arturo » Fri Nov 08, 2002 2:35 am

Heh, it's so frustrating to know about an option that ends up doing very close to nothing. Makes me feel like I'm doing something wrong. :o I really do appreciate the help though. I'll have to start reading Doom9's articles more often.
User avatar
(NERD Studios) Arturo
 
Joined: 19 Jul 2002
Location: Stillwater, OK


Return to Video Software Help Archive

Who is online

Users browsing this forum: No registered users and 1 guest